CN104270312A - Relay path allocation system and method supporting flow optimization and application awareness - Google Patents

Abstract

The invention provides a relay path allocation system and method supporting flow optimization and application awareness and belongs to the technical field of network communication. Bottom layer network information provided by an application layer flow optimization server is utilized, a control server carries out effective organization on a relay overlapping network formed by relay servers according to the bottom layer network topological information, and a relay overlapping network topological model has the characteristics that scalability is high and the relay overlapping network topological model can perceive a bottom layer physical network; the control server obtains multiple types of transmission performance of the relay overlapping network by using the detection performance of the relay servers, thereby having the capacity for application awareness; routing expenses, transmission performance, load rates and other factors are comprehensively considered in a path cost evaluation method for relay paths, so that the relay path allocation system and method have excellent performance in the aspect of selecting high-quality relay paths and can flexibly balance loads among the relay servers.

Description

Support relay route distribution system and the method for flow optimization and application perception

Technical field

The invention belongs to network communication technology field, be specifically related to a kind of relay route distribution system and the method for supporting flow optimization and application perception.

Background technology

In traditional IP, transfer of data depends on the default route path of network layer.But default route path is not optimum under many circumstances, this is because Internet Service Provider selects to be suitable for the path of local policy demand usually, instead of selects the path with top performance.Lot of research shows, and the transmission mode based on application layer overlapping network can provide end-to-end transmission performance more better than default web route transmission pattern.In the transmission mode based on application layer overlapping network, how effectively tissue provides the via node of relay services and selects the relay route of high-quality to be the key point that application layer route plays validity.If relay route has poor transmission performance, the problems such as Consumer's Experience decline end to end, core-network resources waste will be caused, even cause performance more worse than default route path.

Overlap application much popular at present produces a large amount of cross-domain flow rate usually, these cross-domain flow rate cause some hot-spot link or domain-to-domain link to occur congested, except reducing the experience of telex network end to end, also materially increase network operation cost.One of overlapping application major reason consuming Internet resources excessively is the mismatch between overlapping network and bearer network.How perception bearer network situation is the key technology place of application layer traffic transmission optimization scheme.According to the method for different perception bearer network situations, application layer traffic transmission optimization scheme is divided into three kinds: 1) topology estimation is carried out in overlapping application voluntarily, is mainly peer distribution coordinate by detection mode thus estimates the propinquity between node; 2) Virtual network operator is cooperated with application service provider, and Virtual network operator initiatively provides network topological information to application service provider; 3) network topological information of overlapping application use third party measurement, a lot of research and project propose the method for measurement and acquisition network topology at present.All there is the inaccurate problem of network topological information of acquisition in the first and the third method.In the second approach, because Virtual network operator is known quite well own net, therefore, it is possible to provide bearer network information more accurately, and according to the demand amendment Internet resources consumption mode of self strategy, thus both sides' optimization of bearer network and distributed overlay application performance can be reached.

In the transmission mode based on application layer overlapping network, relay route assignment problem needs comprehensive considering various effects, first, different application types may have different transmission demands to relay route, such as, high reliability application requires that relay route has higher path stability and less packet loss; Real-time application requires that relay route has less transmission delay, delay jitter etc.The second, relay route is distributed will take into full account mating between bottom physical network topology, uses bottom-layer network resource rationally and effectively, reduces or avoid the effective consumption of bandwidth resources between territory in the whole network.3rd, relay route distributes transmission performance and the routing cost that not only will consider relay route, also to consider the transmittability of Relay Server itself, while Appropriate application Internet resources, accomplish the flow optimization between territory and between Relay Server and load balancing.As can be seen here, the distribution of relay route involves the organizational form of relaying overlapping network, the many factors such as evaluation, application load transmission demand of path transmission quality.Application number be 201410196986.4 Chinese patent application " a kind of Multipath Transmission system and method based on application layer relay " propose a kind of general Multipath Transmission pattern based on application layer overlapping network, utilize the application layer relay service system be made up of Control Server and Relay Server to set up Multipath Transmission condition end to end.But how effectively to organize Relay Server also for transfer of data distributes the relay route of high-quality end to end, namely this Chinese patent application of relay route assignment problem does not provide corresponding technical scheme.

Summary of the invention

For the deficiency that prior art exists, the invention provides a kind of relay route distribution system and the method for supporting flow optimization and application perception.

Technical scheme of the present invention:

Support a relay route distribution system for flow optimization and application perception, comprising: application layer traffic Optimized Service device, Relay Server and Control Server;

Network Information Organization for obtaining the network information, and is become the form of network mapping table and expense mapping table by described application layer traffic Optimized Service device; Described network mapping table comprises the grouping of a series of network site; The grouping of described network site comprises relam identifier, network address prefix; Described network address prefix represents one or more subnet or represents one or more metropolitan area network or represent one or more autonomous system; Described expense mapping table comprises the routing cost between the grouping of network site; Described routing cost has general meaning, is hop count or route cost;

Described Relay Server is used for execution performance detection process, obtains the transmission performance between itself and other Relay Server, and by the transmission performance report between itself and other Relay Server to Control Server; Described transmission performance comprises time delay and packet loss;

Described Control Server is used for reading network mapping table and expense mapping table from application layer traffic Optimized Service device, according to network mapping table, Relay Server is mapped to network site grouping, the bearer network topology information of Relay Server according to bottom is organized; Obtain the transmission performance between Relay Server from Relay Server, calculate the transmission performance between the grouping of network site according to the transmission performance between network mapping table and Relay Server; Obtain the load factor of Relay Server from Relay Server, the load factor according to network mapping table and Relay Server calculates the territory load factor of dividing into groups network site; According to territory load factor and the relay route allocation request message of the transmission performance between expense mapping table, network site grouping, network site grouping, optimal scheme relay route;

Described Control Server comprises application layer traffic and optimizes client modules, repeater management server module, relaying overlapping network administration module and relay route administration module;

Described application layer traffic is optimized client modules and is used for periodically reading network mapping table and expense mapping table from application layer traffic Optimized Service device;

Described repeater management server module: (a) is in Relay Server registration or connect in Keep Alive procedure, select one group of Relay Server, the address information of selected Relay Server is passed to this Relay Server, transmission performance between obtaining itself and other Relay Server from this Relay Server, and be stored in performance information table; The performance information of (b) inquiry Relay Server and load information, and be stored in relay server information table; Described performance information comprises processor ability, storage capacity and the network bandwidth; Described load information comprises relay route number, processor utilization rate and network bandwidth utilization rate;

Described relaying overlapping network administration module: the Relay Server of each registration, according to network mapping table, is mapped to a network site grouping by (a); B (), according to network mapping table and performance information table, calculate the transmission performance between the grouping of network site, and be stored in corresponding capabilities map table, described capabilities map table is divided into time delay mapping table and packet loss mapping table; C (), according to network mapping table and relay server information table, is each network site grouping computational fields load factor, and is stored in the load factor table of territory; D () is according to network mapping table, expense mapping table, territory load factor table and multiple capabilities map table, calculate the optimum territory relay route based on multiple transmission performance between the grouping of any source network position and the grouping of object network site, and be stored in the relay route table of corresponding optimum territory; Described optimum territory relay route table is divided into the optimum territory relay route table of time delay and packet loss optimum territory relay route table; Described territory relay route is the path of dividing into groups via zero or other network site multiple between the grouping of source network position and the grouping of object network site, and the relam identifier of being divided into groups by network site forms;

Described relay route administration module is used for after receiving relay route allocation request message, generates optimum IP relay route according to optimum territory relay route table; Described IP relay route is by the path via zero or multiple Relay Server between source and destination, is made up of IP address.

Further, described application layer traffic Optimized Service device obtains the mode of the network information, comprises Routing Protocol, static configuration strategy and dynamic network information.

Support a relay route distribution method for flow optimization and application perception, the support flow optimization described in employing realizes with the relay route distribution system of application perception, and concrete steps are as follows:

1) in Relay Server registration or connection Keep Alive procedure, select one group of Relay Server, the address information of selected Relay Server is passed to this Relay Server, transmission performance between obtaining itself and other Relay Server from this Relay Server, and be stored in performance information table;

2) inquire about performance information and the load information of Relay Server, and be stored in relay server information table; Described performance information comprises processor ability, storage capacity and the network bandwidth; Described load information comprises relay route number, processor utilization rate and network bandwidth utilization rate;

3) periodically network mapping table and expense mapping table is read from application layer traffic Optimized Service device;

4) according to network mapping table, the Relay Server of each registration is mapped to a network site grouping;

5) according to network mapping table and performance information table, calculate the transmission performance obtained between the grouping of network site, and be stored in corresponding capabilities map table; Described capabilities map table is divided into time delay mapping table and packet loss mapping table;

6) according to network mapping table and relay server information table, be each network site grouping computational fields load factor, and be stored in the load factor table of territory;

7) according to network mapping table, expense mapping table, territory load factor table and multiple capabilities map table, calculate the optimum territory relay route based on multiple transmission performance obtained between the grouping of any source network position and the grouping of object network site, and be stored in the relay route table of corresponding optimum territory; Described optimum territory relay route table is divided into the optimum territory relay route table of time delay and packet loss optimum territory relay route table; Described territory relay route is the path of dividing into groups via zero or other network site multiple between the grouping of source network position and the grouping of object network site, and the relam identifier of being divided into groups by network site forms;

8) after receiving relay route allocation request message, optimum IP relay route is generated according to optimum territory relay route table; Described IP relay route is by the path via zero or multiple Relay Server between source and destination, is made up of IP address.

Further, described step 4) in the Relay Server of each registration is mapped to a network site grouping, mapping method is mated by the network address prefix divided into groups in the IP address of Relay Server and all-network position in network mapping table, adopt the principle of the longest coupling, choose the grouping of the network site corresponding to the longest network address prefix matched.

Further, described step 7) according to network mapping table, expense mapping table, territory load factor table and multiple capabilities map table, calculate the optimum territory relay route based on multiple transmission performance obtained between the grouping of any source network position and the grouping of object network site, method is: for each transmission performance of specifying, for the grouping of arbitrary source network position and the grouping of object network site, calculate the path cost of all territories relay route between the grouping of source network position and the grouping of object network site, using territory relay route minimum for path cost as the optimum territory relay route based on given transmission performance between the grouping of source network position and the grouping of object network site.

Further, described step 7) according to network mapping table, expense mapping table, territory load factor table and multiple capabilities map table, calculate the optimum territory relay route based on multiple transmission performance obtained between the grouping of any source network position and the grouping of object network site, method can also be:

For each transmission performance of specifying, for the grouping of arbitrary source network position and the grouping of object network site, calculate the path cost via all territories relay route of maximum k (k >=1) other network site individual grouping between the grouping of source network position and the grouping of object network site, using territory relay route minimum for path cost as the optimum territory relay route based on given transmission performance between the grouping of source network position and the grouping of object network site.

Further, described path cost comprises path routing cost cost and path transmission performance cost; Described path routing cost cost is the routing cost sum in the relay route of territory between the grouping of adjacent two network sites; Described path transmission performance cost is divided into path delay cost and path packet loss cost; Described path delay cost is the time delay sum in the relay route of territory between the grouping of adjacent two network sites; Described path packet loss cost is

$\mathrm{PPATH}\_{\mathrm{Lossrate}}_{{\mathrm{PID}}_{n_0}->{\mathrm{PID}}_{n_{m+1}}}=1-\underset{0\≤i\≤m}{\mathrm{\Π}}(1-\mathrm{PID}\_{\mathrm{Lossrate}}_{n_i{n}_{i+1}}),$ Wherein, represent the grouping of source network position, represent the grouping of object network site, m represents with between a territory relay route the number of dividing into groups via other network site, represent with between packet loss,

represent with between the path packet loss of a territory relay route.

Or further, described path cost comprises path routing cost cost, path transmission performance cost and path load rate cost; Described path load rate cost is the territory load factor maximum of network site grouping in the relay route of territory.

Or further, described path cost comprises path routing cost cost, path transmission performance cost and path relaying cost; Described path relaying cost and territory relay route via network site number of dividing into groups be directly proportional.

Or further, described path cost comprises path routing cost cost, path transmission performance cost, path load rate cost and path relaying cost;

Further, described step 8) in after receiving relay route allocation request message, generate optimum IP relay route according to optimum territory relay route table, method is:

(1) from relaying path allocation request message, application type, source IP address and object IP address is extracted;

(2) according to network mapping table, according to longest match principle, source IP address and object IP address are mapped to the grouping of source network position and the grouping of object network site respectively;

(3) determine preferred transmission performance according to application type, find corresponding optimum territory relay route table according to preferred transmission performance;

(4) from the relay route table of optimum territory, obtain the optimum territory relay route between the grouping of source network position and the grouping of object network site;

(5) if optimum territory relay route is not divided into groups via other network site except the grouping of source network position and the grouping of object network site, territory load factor is selected less from both the grouping of source network position and the grouping of object network site, Stochastic choice one maps to the Relay Server of this network site grouping, and optimum IP relay route is made up of source IP address, the IP address of selected Relay Server, object IP address;

(6) if optimum territory relay route is also divided into groups via other network sites one or more except the grouping of source network position and the grouping of object network site, for each via the grouping of other network site, Stochastic choice one maps to the Relay Server of this network site grouping, and optimum IP relay route is made up of source IP address, the IP address of one or more selected Relay Server, object IP address.

Beneficial effect of the present invention: utilize the bottom-layer network information that application layer traffic Optimized Service device provides, the relaying overlapping network that Relay Server forms by Control Server is effectively organized according to bottom-layer network topology information, make relaying overlay network topology model have scalability strong, with the feature of bottom physical network phase perception; Utilize the detection performance between Relay Server, Control Server obtains the multinomial transmission performance of relaying overlapping network, thus has the ability of application perception; The path cost evaluation method of relay route considers the many factors such as routing cost, transmission performance, load factor, the present invention is made not only in selection high-quality relay route, to have comparatively dominance energy, simultaneously again can load neatly between balanced Relay Server.

Accompanying drawing explanation

Fig. 1 is the support flow optimization of one embodiment of the present invention and the relay route distribution system configuration schematic diagram of application perception;

Fig. 2 is the relaying overlay network topology model of the Relay Server composition of one embodiment of the present invention;

Fig. 3 is the support flow optimization of one embodiment of the present invention and the relay route distribution method flow chart of application perception;

Fig. 4 is the generation optimum IP relay route method flow diagram of one embodiment of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is elaborated.

The support flow optimization of present embodiment and the relay route distribution system of application perception, as shown in Figure 1, comprise application layer traffic Optimized Service device, Control Server and Relay Server.

Network Information Organization, for obtaining the network information, is become the form of network mapping table and expense mapping table by application layer traffic Optimized Service device.Application layer traffic Optimized Service device obtains the network information by various ways, such as Routing Protocol, static configuration strategy, dynamic network information, or even passes through the third party content provider of external interface.

Network mapping table comprises the grouping of a series of network site, and network site grouping comprises relam identifier, network address prefix, and network address prefix can represent one or more subnet, metropolitan area network and autonomous system.In real network, a lot of terminal is close to over network connections very much, is such as in the terminal of same company, colleges and universities.These adjacent terminals are abstracted into a network site grouping, larger scalability can be obtained, do not lose important network topological information simultaneously.Divided into groups by network site, application layer traffic Optimized Service device provides the network terminal of topology Network Based or other network attribute to be polymerized in a kind of mode indirectly had nothing to do with network.

Expense mapping table comprises the routing cost between the grouping of network site.Routing cost refers to the end-to-end cost of unit discharge from source to object, has general meaning, and Virtual network operator can carry out internal calculation according to self strategy, such as can according to the physical distance between network, hop count or route cost etc.Application layer traffic Optimized Service device utilizes the priority between the form formulation network site grouping of routing cost, to reach the object of amendment Internet resources consumption mode.

The network information is divided into network mapping table and expense mapping table by application layer traffic Optimized Service device, its benefit is, network topology and network state two category information can upgrade on Different time scales, namely reflect that the network mapping table of network topological information may not change in a long time, and reflect that the expense mapping table of network dynamic state information may change within a short period of time.

Relay Server in present embodiment adopts application number to be the relay services transponder that in the Chinese patent application " a kind of Multipath Transmission system and method based on application layer relay " of 201410196986.4, claim 4 is recorded.Relay Server is according to response message or the list of connection keep-alive message structure Relay Server of succeeding in registration, and execution performance detection process.Covering link between Relay Server is made up of multiple physical link usually; Except covering flow, non-covered flow also will use identical physical link.Therefore, Relay Server can not control or the Internet resources of managing I P layer, can only depend on detection mechanism and obtain the performance information covering link between other Relay Server.Utility detection process, Relay Server can obtain the dynamic transmission performance covering link between other Relay Server, and the transmission performance information of detection is reported to Control Server via connection keep-alive message.

Relay Server needs one group to detect operator for evaluating the transmission performance covering link between other Relay Server.The Main Function of these detection operators is exactly provide according to corresponding transmission performance tolerance the quantitative values that covers link quality.These quantitative values are relative, and other quantitative values only for obtaining with same detection operator compares.Under default situations, Relay Server supports that two detect operator: time delay detection operator and packet loss detection operator, be respectively used to detect the time delay and the packet loss that cover link between Relay Server.In actual deployment, new performance metric can be set up according to the concrete transmission demand of upper layer application, such as throughput, delay jitter etc., even define new performance metric for certain class upper layer application, supported to detect operator accordingly for Relay Server expands by these needs.Provide a kind of realization example of time delay detection operator and packet loss detection operator below.

Time delay detection operator, periodically to each Relay Server forward delay interval detection packet in Relay Server list, calculates and obtains the associated round-trip delay sample covering link.The Delay Estima-tion value covering link can adopt the exponential weighting slip average of round trip delay time sample value.R_Delay _ijrepresent Relay Server R _iand R _jbetween cover the time delay assessed value of link, calculate according to formula (1):

R_Delay _ij＝αR_Delay _ij+(1-α)New_Sample _ij (1)

Wherein, New_Sample _ijrepresent the new round-trip delay sample obtained, parameter alpha is slip weights, and default value is 0.9, represents Delay Estima-tion value R_Delay _ij10% be taken from up-to-date round-trip delay sample value.

Packet loss detection operator periodically sends continuous print n to each Relay Server in Relay Server list ₁(default value is 10) individual packet loss detection packet, calculates and obtains the associated packet loss sample covering link.What the packet loss sample value calculated provided is two-way drop probabilities, can estimate to obtain unidirectional drop probabilities under the prerequisite of supposition packet loss symmetry.The packet loss estimated value covering link adopts up-to-date n ₂the mean value computation of (default value is 10) individual packet loss sample value.

Control Server in present embodiment adopts application number to be the relay services controller that in the Chinese patent application " a kind of Multipath Transmission system and method based on application layer relay " of 201410196986.4, claim 3 is recorded.Control Server reads network mapping table and expense mapping table from application layer traffic Optimized Service device, according to network mapping table, Relay Server is mapped to network site grouping, the bearer network topology information of Relay Server according to bottom is organized.By network mapping table and expense mapping table, Control Server can know the network information of bottom-layer network architecture, thus has the ability as having the user agent of data transfer demands to distribute bottom-layer network perception and having the relay route of Minimal routing expense.Control Server obtains the transmission performance between Relay Server from Relay Server, calculates the transmission performance obtained between the grouping of network site according to the transmission performance between network mapping table and Relay Server.Control Server obtains the load factor of Relay Server from Relay Server, and the load factor according to network mapping table and Relay Server calculates the territory load factor obtaining network site grouping.Transmission performance between dividing into groups according to expense mapping table, network site, the territory load factor of network site grouping and the relay route allocation request message coming from user agent or third party's signal server received, Control Server can distribute optimum relay route for the transfer of data between user agent.

Divide according to function, Control Server comprises application layer traffic and optimizes client modules, repeater management server module, relaying overlapping network administration module and relay route administration module.

Application layer traffic is optimized client modules and is periodically obtained network mapping table and expense mapping table from application layer traffic Optimized Service device.Application layer traffic optimizes client modules needs the address knowing application layer traffic Optimized Service device.The mode that application layer traffic optimizes client modules acquisition application layer traffic Optimized Service device address can have a lot, such as adopts DNS query to obtain HTTP (S) URI of application layer traffic Optimized Service device.

The registration of Relay Servers all in repeater management server module management and maintenance field and state information: the Relay Server sending performance query messages after registration, inquire about its performance information, comprise processor ability, storage capacity and the network bandwidth etc.; Periodically send STATUS ENQUIRY message to available Relay Server, inquire about the load information that it is current, comprise relay route number, processor utilization rate and network bandwidth utilization rate etc.The performance of Relay Server and load information are stored in relay server information table by repeater management server module.In Relay Server registration or connect in Keep Alive procedure, repeater management server module in the grouping of each network site at random or according to certain policy selection n ₃(n ₃>=1) individual registered Relay Server, and its address information is passed to Relay Server.Repeater management server module obtains the transmission performance between this Relay Server and other Relay Server from Relay Server, and is stored in performance information table.

Relaying overlapping network administration module is according to network mapping table, and the Relay Server of each registration is mapped to a network site grouping, the grouping of this network site can cover the IP address of this Relay Server.Mapping method is mated by the network address prefix divided into groups in the IP address of Relay Server and all-network position in network mapping table, the network address prefix matched may have multiple, adopt the principle of the longest coupling, choose the grouping of the network site corresponding to the longest network address prefix matched.Obviously, the grouping of each network site may comprise multiple Relay Server.In this manner, Relay Server divides according to network site grouping by Control Server, and network site grouping represents one group of geographical position and closes on and the similar address of routing policy, therefore, the bearer network topology information of Relay Server according to bottom is organized by Control Server, geographical position to be closed on and the similar Relay Server of routing policy is divided into one group, can perform identical relay route allocation strategy to it, the relaying overlay network topology structure that Relay Server is formed can adapt to the bearer network of bottom well.

As shown in Figure 2, relaying overlapping network has been organized into two-layer model by Control Server.The discrete networks that lower floor is made up of Relay Server, Control Server and each relay server, safeguard the information such as the performance of single Relay Server, load.Full mesh (full mesh) network that upper strata is made up of network site packet node, by belong to same network site grouping multiple Relay Servers be abstracted into individual node, Control Server safeguard any two network site packet nodes between transmission performance information.The number of suppose relay server is N, and the number of network site packet node is M, obvious M < < N, and Control Server needs the amount of information safeguarded to be O (N+M ²).Along with the increase of userbase in system, manager may set up new Relay Server for sharing the overlapping load of increase, at this moment Control Server only needs the maintenance information linearly increasing lower floor's Relay Server, and upper layer network position packet node number increases slowly under normal circumstances.Therefore, the two-layer topological model of relaying overlapping network has good network scalability.

Performance information between the grouping of relaying overlapping network administration module maintaining network position.Transmission performance between the grouping of two network sites can with the multiple Relay Servers in these two network sites groupings between the average of respective transmissions performance calculate.Under default situations, two transmission performance tolerance between the grouping of relaying overlapping network administration module management maintenance network site: time delay and packet loss, be stored in time delay mapping table and packet loss mapping table respectively.PID_Delay _ijrepresent network site group PID _iand PID _jbetween time delay assessed value, PID_Lossrate _ijrepresent network site group PID _iand PID _jbetween packet loss estimated value, respectively according to following formulae discovery:

$\mathrm{PID}\_{\mathrm{Delay}}_{\mathrm{ij}}={\mathrm{avg}}_{R_k\&Element;{\mathrm{PID}}_i,R_m\&Element;{\mathrm{PID}}_j}(R\_{\mathrm{Delay}}_{\mathrm{km}})---\left(2\right)$

$\mathrm{PID}\_{\mathrm{Lossrate}}_{\mathrm{ij}}={\mathrm{avg}}_{R_k\&Element;{\mathrm{PID}}_i,R_m\&Element;{\mathrm{PID}}_j}(R\_{\mathrm{Lossrate}}_{\mathrm{km}})---\left(3\right)$

Wherein, R_Delay _kmrepresent Relay Server R _kand R _mbetween Delay Estima-tion value, R_Lossrate _kmrepresent Relay Server R _kand R _mbetween packet loss estimated value.

According to relay server information table and network mapping table, relaying overlapping network administration module is each network site grouping computational fields load factor, and leaves in the load factor table of territory.PID_Loadrate _irepresent network site group PID _iterritory load factor, according to PID _ithe load factor average of all Relay Servers comprised calculates, that is:

$\mathrm{PID}\_{\mathrm{Loadrate}}_i={\mathrm{avg}}_{R_k\&Element;{\mathrm{PID}}_i}(R\_{\mathrm{Loadrate}}_k)---\left(4\right)$

Wherein, R_Loadrate _krepresent Relay Server R _kload factor.

According to network mapping table, expense mapping table, territory load factor table and multiple capabilities map table, relaying overlapping network administration module calculates the optimum territory relay route based on multiple transmission performance obtained between the grouping of any source network position and the grouping of object network site, and leaving in the relay route table of corresponding optimum territory, optimum territory relay route table is divided into the optimum territory relay route table of time delay and packet loss optimum territory relay route table.Territory relay route is the path of dividing into groups via zero or other network site multiple between the grouping of source network position and the grouping of object network site, and the relam identifier of being divided into groups by network site forms.After network mapping table, expense mapping table, territory load factor table or corresponding capabilities map table change, relaying overlapping network administration module regenerates optimum territory relay route table.

After receiving the relay route allocation request message coming from user agent or third party's signal server, relay route administration module generates optimum IP relay route according to optimum territory relay route table.IP relay route is by the path via zero or multiple Relay Server between source and destination, is made up of IP address.

The support flow optimization of present embodiment and the relay route distribution method of application perception, as shown in Figure 3, specifically comprise the steps:

1) in Relay Server registration or connection Keep Alive procedure, select one group of Relay Server, the address information of selected Relay Server is passed to this Relay Server, transmission performance between obtaining itself and other Relay Server from this Relay Server, and be stored in performance information table; Transmission performance between Relay Server comprises time delay and packet loss, and in actual deployment, can set up new performance metric according to the concrete transmission demand of upper layer application, such as throughput, delay jitter etc., even define new performance metric for certain class upper layer application;

2) inquire about performance information and the load information of Relay Server, and be stored in relay server information table; Performance information comprises processor ability, storage capacity and the network bandwidth; Load information comprises relay route number, processor utilization rate and network bandwidth utilization rate;

3) periodically network mapping table and expense mapping table is read from application layer traffic Optimized Service device;

4) according to network mapping table, the Relay Server of each registration is mapped to a network site grouping, mapping method is mated by the network address prefix divided into groups in the IP address of Relay Server and all-network position in network mapping table, adopt the principle of the longest coupling, choose the grouping of the network site corresponding to the longest network address prefix matched;

5) according to network mapping table and performance information table, calculate the transmission performance obtained between the grouping of network site, and be stored in corresponding capabilities map table, capabilities map table is divided into time delay mapping table and packet loss mapping table;

6) according to network mapping table and relay server information table, be each network site grouping computational fields load factor, and be stored in the load factor table of territory;

7) according to network mapping table, expense mapping table, territory load factor table and multiple capabilities map table, calculate the optimum territory relay route based on multiple transmission performance obtained between the grouping of any source network position and the grouping of object network site, and be stored in the relay route table of corresponding optimum territory; Optimum territory relay route table is divided into the optimum territory relay route table of time delay and packet loss optimum territory relay route table; Territory relay route is the path of dividing into groups via zero or other network site multiple between the grouping of source network position and the grouping of object network site, and the relam identifier of being divided into groups by network site forms;

For each transmission performance of specifying, for the grouping of arbitrary source network position and the grouping of object network site, relaying overlapping network administration module calculates the path cost of all territories relay route between the grouping of source network position and the grouping of object network site, using territory relay route minimum for path cost as the optimum territory relay route based on given transmission performance between the grouping of source network position and the grouping of object network site.Experimental result shows, the overlay path via single via node can provide the performance close with the optimum overlay path via multiple via node in most cases.Therefore, when calculating optimum territory relay route, relaying overlapping network administration module can limit at most via k (k >=1) other network site individual grouping between the grouping of source network position and the grouping of object network site, on this basis using territory relay route minimum for path cost as the optimum territory relay route based on given transmission performance between the grouping of source network position and the grouping of object network site.

Relay route assigning process should consider factor in many ways, comprise routing cost, transmission performance, network site grouping or the load information of Relay Server, can hold concurrently routing policy, apply preferred performance metric, can avoid again indivedual Relay Server or certain network site packet payload overweight.Therefore, when calculating optimum territory relay route, the computational methods of the path cost of territory relay route are most important.

Path cost comprises path routing cost cost and path transmission performance cost. represent the grouping of source network position divide into groups with object network site between a territory relay route, $\mathrm{PID}-P_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}=({\mathrm{PID}}_{n_0},{\mathrm{PID}}_{n_1},...,{\mathrm{PID}}_{n_m},{\mathrm{PID}}_{n_{m+1}}),$ M (m>=0) is territory relay route via other network site grouping number. representative domain relay route path cost, be defined as following formula:

$\begin{array}{c}{\mathrm{PCost}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}=a_1\&times;\mathrm{PPATH}\_{\mathrm{Route}\cos t}_{{\mathrm{PID}}_{n_0}->{\mathrm{PID}}_{n_{m+1}}}\\ +a_2\&times;\mathrm{PPATH}\_{\mathrm{QoS}}_{{\mathrm{PID}}_{n_0}->{\mathrm{PID}}_{n_{m+1}}}\end{array}---\left(5\right)$

Wherein, a ₁and a ₂for configuration parameter, meet a ₁+ a ₂=1; representative domain relay route path routing cost cost, be the routing cost sum in the relay route of territory between the grouping of adjacent two network sites; representative domain relay route path transmission performance cost.

Route network can not be optimized multiple performance metric usually simultaneously, and such as, a path with less time delay may have larger packet loss.Therefore, only based on single performance metric calculation path transmission performance in implementation process, namely path transmission performance cost can be path delay cost, also can be path packet loss cost, or specifically transmits according to upper layer application other transmission performance tolerance that demand newly introduces.Path delay cost is the time delay sum in the relay route of territory between the grouping of adjacent two network sites.The packet loss performance metric in path obeys multiplication relation, representative domain relay route path packet loss cost, according to following formulae discovery:

$\mathrm{PPATH}\_{\mathrm{Lossrate}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}=1-\underset{0\&le;i\&le;m}{\mathrm{\&Pi;}}(1-\mathrm{PID}\_{\mathrm{Lossrate}}_{n_i{n}_{i+1}})---\left(6\right)$

Except path routing cost cost and path transmission performance cost, path cost can also comprise path load rate cost, that is:

$\begin{array}{c}{\mathrm{PCost}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}=a_1\&times;\mathrm{PPATH}\_{\mathrm{Route}\cos t}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}\\ +a_2\&times;\mathrm{PPATH}\_{\mathrm{QoS}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}\\ +a_3\&times;\mathrm{PPATH}\_{\mathrm{Loadrate}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}\end{array}---\left(7\right)$

Wherein, a ₁, a ₂and a ₃for configuration parameter, meet a ₁+ a ₂+ a ₃=1; representative domain relay route path load rate cost, be in the relay route of territory network site grouping territory load factor maximum.

Except path routing cost cost and path transmission performance cost, path cost can also comprise path relaying cost.Reason is, Relay Server provides in the process of relay function the resource such as disposal ability, the network bandwidth that can consume self, nor invariably can introduce extra process time delay, thus adds the transmission delay that data arrive receiving terminal.The computing formula of path cost is as follows:

$\begin{array}{c}{\mathrm{PCost}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}=a_1\&times;\mathrm{PPATH}\_{\mathrm{Route}\cos t}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}\\ +a_2\&times;\mathrm{PPATH}\_{\mathrm{QoS}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}\\ +a_3\&times;n\&times;\mathrm{RC}\end{array}---\left(8\right)$

Wherein, a ₁, a ₂and a ₃for configuration parameter, meet a ₁+ a ₂+ a ₃=1; N is territory relay route via provide relay services network site grouping number, when m>=1, n=m, as m=0, n=1; RC represents the relaying cost that Relay Server provides relay function to pay for a data flow; That is, path relaying cost and territory relay route via the network site of the relay services number of dividing into groups that provides be directly proportional.

Except path routing cost cost and path transmission performance cost, path cost can also comprise path load rate cost and path relaying cost simultaneously, that is:

$\begin{array}{c}{\mathrm{PCost}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}=a_1\&times;\mathrm{PPATH}\_{\mathrm{Route}\cos t}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}\\ +a_2\&times;\mathrm{PPATH}\_{\mathrm{QoS}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}\\ +a_3\&times;\mathrm{PPATH}\_{\mathrm{Loadrate}}_{{\mathrm{PID}}_{n_0}\&RightArrow;{\mathrm{PID}}_{n_{m+1}}}\\ +a_4\&times;n\&times;\mathrm{RC}\end{array}---\left(9\right)$

Wherein, a ₁, a ₂, a ₃and a ₄for configuration parameter, meet a ₁+ a ₂+ a ₃+ a ₄=1.

8) after receiving the relay route allocation request message coming from user agent or third party's signal server, optimum IP relay route is generated according to optimum territory relay route table; IP relay route is by the path via zero or multiple Relay Server between source and destination, is made up of IP address;

Generate optimum IP relay route according to optimum territory relay route table, as shown in Figure 4, specifically comprise the steps:

(1) from relaying path allocation request message, application type, source IP address src_ip and object IP address dst_ip is taken out;

(2) according to network mapping table, according to longest match principle, source IP address src_ip and object IP address dst_ip is mapped to the grouping of source network position respectively divide into groups with object network site

(3) determine preferred transmission performance according to application type, find corresponding optimum territory relay route table according to preferred transmission performance;

(4) from the relay route table of optimum territory, obtain the grouping of source network position divide into groups with object network site between optimum territory relay route

(5) if optimum territory relay route except the grouping of source network position divide into groups with object network site outside do not divide into groups via other network site, divide into groups from source network position divide into groups with object network site select territory load factor in both less, Stochastic choice one maps to the Relay Server R of this network site grouping _selected, optimum IP relay route is by source IP address, selected Relay Server R _selectediP address, object IP address composition.

(6) if optimum territory relay route except the grouping of source network position divide into groups with object network site outside also via other network site group PID one or more _mid, for each via other network site group PID _mid, Stochastic choice one maps to the Relay Server of this network site grouping, and optimum IP relay route is made up of source IP address, the IP address of one or more selected Relay Server, object IP address.

Although the foregoing describe the specific embodiment of the present invention, the those skilled in the art in this area should be appreciated that these only illustrate, can make various changes or modifications, and do not deviate from principle of the present invention and essence to these execution modes.Scope of the present invention is only defined by the appended claims.

Claims (9)

1. support a relay route distribution system for flow optimization and application perception, it is characterized in that: comprising: application layer traffic Optimized Service device, Relay Server and Control Server;

Network Information Organization for obtaining the network information, and is become the form of network mapping table and expense mapping table by described application layer traffic Optimized Service device; Described network mapping table comprises the grouping of a series of network site; The grouping of described network site comprises relam identifier, network address prefix; Described network address prefix represents one or more subnet or represents one or more metropolitan area network or represent one or more autonomous system; Described expense mapping table comprises the routing cost between the grouping of network site; Described routing cost has general meaning, is hop count or route cost;

Described Relay Server is used for execution performance detection process, obtains the transmission performance between itself and other Relay Server, and by the transmission performance report between itself and other Relay Server to Control Server; Described transmission performance comprises time delay and packet loss;

Described Control Server is used for reading network mapping table and expense mapping table from application layer traffic Optimized Service device, according to network mapping table, Relay Server is mapped to network site grouping, the bearer network topology information of Relay Server according to bottom is organized; Obtain the transmission performance between Relay Server from Relay Server, calculate the transmission performance between the grouping of network site according to the transmission performance between network mapping table and Relay Server; Obtain the load factor of Relay Server from Relay Server, the load factor according to network mapping table and Relay Server calculates the territory load factor of dividing into groups network site; According to territory load factor and the relay route allocation request message of the transmission performance between expense mapping table, network site grouping, network site grouping, optimal scheme relay route;

Described Control Server comprises application layer traffic and optimizes client modules, repeater management server module, relaying overlapping network administration module and relay route administration module;

Described application layer traffic is optimized client modules and is used for periodically reading network mapping table and expense mapping table from application layer traffic Optimized Service device;

Described repeater management server module: (a) is in Relay Server registration or connect in Keep Alive procedure, select one group of Relay Server, the address information of selected Relay Server is passed to this Relay Server, transmission performance between obtaining itself and other Relay Server from this Relay Server, and be stored in performance information table; The performance information of (b) inquiry Relay Server and load information, and be stored in relay server information table; Described performance information comprises processor ability, storage capacity and the network bandwidth; Described load information comprises relay route number, processor utilization rate and network bandwidth utilization rate;

Described relaying overlapping network administration module: the Relay Server of each registration, according to network mapping table, is mapped to a network site grouping by (a); B (), according to network mapping table and performance information table, calculate the transmission performance between the grouping of network site, and be stored in corresponding capabilities map table, described capabilities map table is divided into time delay mapping table and packet loss mapping table; C (), according to network mapping table and relay server information table, is each network site grouping computational fields load factor, and is stored in the load factor table of territory; D () is according to network mapping table, expense mapping table, territory load factor table and multiple capabilities map table, calculate the optimum territory relay route based on multiple transmission performance between the grouping of any source network position and the grouping of object network site, and be stored in the relay route table of corresponding optimum territory; Described optimum territory relay route table is divided into the optimum territory relay route table of time delay and packet loss optimum territory relay route table; Described territory relay route is the path of dividing into groups via zero or other network site multiple between the grouping of source network position and the grouping of object network site, and the relam identifier of being divided into groups by network site forms;

Described relay route administration module is used for after receiving relay route allocation request message, generates optimum IP relay route according to optimum territory relay route table; Described IP relay route is by the path via zero or multiple Relay Server between source and destination, is made up of IP address.

2. the relay route distribution system of support flow optimization according to claim 1 and application perception, it is characterized in that: described application layer traffic Optimized Service device obtains the mode of the network information, comprises Routing Protocol, static configuration strategy and dynamic network information.

3. support a relay route distribution method for flow optimization and application perception, adopt the relay route distribution system of support flow optimization according to claim 1 and application perception to realize, it is characterized in that: concrete steps are as follows:

1) in Relay Server registration or connection Keep Alive procedure, select one group of Relay Server, the address information of selected Relay Server is passed to this Relay Server, transmission performance between obtaining itself and other Relay Server from this Relay Server, and be stored in performance information table;

2) inquire about performance information and the load information of Relay Server, and be stored in relay server information table; Described performance information comprises processor ability, storage capacity and the network bandwidth; Described load information comprises relay route number, processor utilization rate and network bandwidth utilization rate;

3) periodically network mapping table and expense mapping table is read from application layer traffic Optimized Service device;

4) according to network mapping table, the Relay Server of each registration is mapped to a network site grouping;

5) according to network mapping table and performance information table, calculate the transmission performance obtained between the grouping of network site, and be stored in corresponding capabilities map table; Described capabilities map table is divided into time delay mapping table and packet loss mapping table;

6) according to network mapping table and relay server information table, be each network site grouping computational fields load factor, and be stored in the load factor table of territory;

7) according to network mapping table, expense mapping table, territory load factor table and multiple capabilities map table, calculate the optimum territory relay route based on multiple transmission performance obtained between the grouping of any source network position and the grouping of object network site, and be stored in the relay route table of corresponding optimum territory; Described optimum territory relay route table is divided into the optimum territory relay route table of time delay and packet loss optimum territory relay route table; Described territory relay route is the path of dividing into groups via zero or other network site multiple between the grouping of source network position and the grouping of object network site, and the relam identifier of being divided into groups by network site forms;

8) after receiving relay route allocation request message, optimum IP relay route is generated according to optimum territory relay route table; Described IP relay route is by the path via zero or multiple Relay Server between source and destination, is made up of IP address.

4. the relay route distribution method of support flow optimization according to claim 3 and application perception, it is characterized in that: described step 4) in the Relay Server of each registration is mapped to a network site grouping, mapping method is mated by the network address prefix divided into groups in the IP address of Relay Server and all-network position in network mapping table, adopt the principle of the longest coupling, choose the grouping of the network site corresponding to the longest network address prefix matched.

5. the relay route distribution method of support flow optimization according to claim 3 and application perception, it is characterized in that: described step 7) according to network mapping table, expense mapping table, territory load factor table and multiple capabilities map table, calculate the optimum territory relay route based on multiple transmission performance obtained between the grouping of any source network position and the grouping of object network site, computational methods are method A or method B, wherein method A is: for each transmission performance of specifying, for the grouping of arbitrary source network position and the grouping of object network site, calculate the path cost of all territories relay route between the grouping of source network position and the grouping of object network site, using territory relay route minimum for path cost as the optimum territory relay route based on given transmission performance between the grouping of source network position and the grouping of object network site, method B is: for each transmission performance of specifying, for the grouping of arbitrary source network position and the grouping of object network site, calculate the path cost via all territories relay route of maximum k (k >=1) other network site individual grouping between the grouping of source network position and the grouping of object network site, using territory relay route minimum for path cost as the optimum territory relay route based on given transmission performance between the grouping of source network position and the grouping of object network site.

6. the relay route distribution method of support flow optimization according to claim 5 and application perception, is characterized in that: described path cost comprises path routing cost cost and path transmission performance cost; Described path routing cost cost is the routing cost sum in the relay route of territory between the grouping of adjacent two network sites; Described path transmission performance cost is divided into path delay cost and path packet loss cost; Described path delay cost is the time delay sum in the relay route of territory between the grouping of adjacent two network sites; Described path packet loss cost is $\mathrm{PPATH}\_{\mathrm{Lossrate}}_{{\mathrm{PID}}_{n_0}->{\mathrm{PID}}_{n_{m+1}}}=1-\underset{0\&le;i\&le;m}{\mathrm{\&Pi;}}(1-\mathrm{PID}\_{\mathrm{Lossrate}}_{n_i{n}_{i+1}}),$ Wherein, represent the grouping of source network position, represent the grouping of object network site, m represents with between a territory relay route the number of dividing into groups via other network site, represent with between packet loss, represent with between the path packet loss of a territory relay route.

7. the relay route distribution method of support flow optimization according to claim 6 and application perception, is characterized in that: described path cost also comprises path load rate cost; Described path load rate cost is the territory load factor maximum of network site grouping in the relay route of territory.

8. the relay route distribution method of the support flow optimization according to claim 6 or 7 and application perception, is characterized in that: described path cost also comprises path relaying cost; Described path relaying cost and territory relay route via network site number of dividing into groups be directly proportional.

9. the relay route distribution method of support flow optimization according to claim 3 and application perception, it is characterized in that: described step 8) in after receiving relay route allocation request message, generate optimum IP relay route according to optimum territory relay route table, method is:

(1) from relaying path allocation request message, application type, source IP address and object IP address is extracted;

(2) according to network mapping table, according to longest match principle, source IP address and object IP address are mapped to the grouping of source network position and the grouping of object network site respectively;

(3) determine preferred transmission performance according to application type, find corresponding optimum territory relay route table according to preferred transmission performance;

(4) from the relay route table of optimum territory, obtain the optimum territory relay route between the grouping of source network position and the grouping of object network site;

(5) if optimum territory relay route is not divided into groups via other network site except the grouping of source network position and the grouping of object network site, territory load factor is selected less from both the grouping of source network position and the grouping of object network site, Stochastic choice one maps to the Relay Server of this network site grouping, and optimum IP relay route is made up of source IP address, the IP address of selected Relay Server, object IP address;

(6) if optimum territory relay route is also divided into groups via other network sites one or more except the grouping of source network position and the grouping of object network site, for each via the grouping of other network site, Stochastic choice one maps to the Relay Server of this network site grouping, and optimum IP relay route is made up of source IP address, the IP address of one or more selected Relay Server, object IP address.